Muscle Memory of Exercise Optimizes Mitochondrial Metabolism to Support Skeletal Muscle Growth

Abstract

Exercise protects against age-related declines in skeletal muscle mass and function while improving overall health. Exercise can also prime long-term muscle health to enhance adaptations upon exercise retraining, a phenomenon termed muscle memory that remains largely understudied. To assess how prior endurance training elicits a lasting metabolic memory in skeletal muscle, we utilized C57BL/6 mice fed either a control (CD) or obesogenic diet (HFD) that underwent 4-week training, detraining, and retraining periods. Our results show that exercise retraining attenuated weight gain and potentiated muscle growth, even with reduced voluntary running volumes. Training increased fiber size (fCSA), which disappeared with detraining and was recovered with retraining regardless of diet, pointing to a glycolytic-to-oxidative fiber shift. Transcriptomic analysis (bulk RNA-seq) of the retrained muscle revealed a robust enhancement of mitochondrial oxidative phosphorylation (OxPhos) and mitoribosomal genes, paralleled by increases in OxPhos protein complex IV levels, higher long-chain fatty acid oxidative capacity (ACADL), and sustained citrate synthase activity 1 week after retraining, reinforcing the optimization of mitochondrial metabolism. While transcriptomic evidence revealed a major overlap between HFD- and CD-fed mice, discrepancies in protein abundance emerged, which point to an intricate regulation of mitochondrial programming that supports the muscle memory of growth. Our study identifies common and selective mechanisms by which the muscle memory of exercise overrides dietary challenges and promotes fiber hypertrophy, offering insight into potential mechanisms to leverage to promote healthy aging.

NEW & NOTEWORTHY

Here we provide evidence that exercise memory in skeletal muscle fine-tunes mitochondrial metabolism to respond to dietary challenges and support muscle growth. Using physiological, RNA sequencing, and biochemical approaches, we show that exercise retraining optimizes mitochondrial metabolism to increase fatty acid oxidative capacity. These findings enhance our understanding of how prior exercise primes muscle for enhanced adaptations, offering insights into strategies to promote healthy aging.

Graphical Abstract

INTRODUCTION

Skeletal muscle plays a crucial role in supporting overall health, particularly as we age. In that sense, preserving or increasing muscle mass is crucial for reducing frailty, preserving mobility, and improving metabolic health (1). In addition to facilitating movement, skeletal muscle plays a major role in energy homeostasis, accounting for the majority of glucose disposal and contributes to the prevention of chronic diseases such as type 2 diabetes and cardiovascular diseases (2). Nevertheless, aging accelerates the loss of muscle mass and strength, a condition known as sarcopenia, which diminishes the quality of life in elderly individuals (3). Thus, identifying therapies that can delay and prevent the onset of muscle functional decline is of the utmost importance.

Exercise is a powerful lifestyle intervention that can protect muscles against age- and disease-related decline while increasing their metabolic, contractile, and endocrine capacity (1, 4–6). Endurance exercise effectively improves skeletal muscle health and function by enhancing aerobic capacity and mitochondrial function (7); however, whether these benefits can persist long-term is only beginning to be explored. Colloquially known as ‘muscle memory,’ this concept describes the body’s ability to perform learned tasks with ease, even after extended periods without practice. Recent evidence has shed light on the mechanistic underpinnings of the muscle memory utilizing cycles of endurance- or resistance-based exercise training, unloading, and exercise retraining. Myonuclear permanence, or the preservation of myonuclei during unloading following a hypertrophic stimulus, has been proposed as a driver of the muscle memory (8). Nevertheless, conflicting evidence in mice shows detraining reduces myonuclei density (9), and a meta-analysis of exercise training in humans indicated that myonuclei may not be retained (10), thus pointing to important discrepancies. Other mechanisms involve alterations to the epigenetic landscape, pointing to microRNAs (miRNAs) and DNA methylation as putative targets. Utilizing a progressive weighted wheel running (PoWeR) of endurance, miR-1 was reduced with exercise and remained reduced throughout the detraining period (11), which might impact energy metabolism through regulation of glycolytic flux and pyruvate metabolism (12). On the other hand, alterations in DNA methylation have been linked to muscle hypertrophy, particularly genes involved in the mTOR and autophagy pathways (13). These changes to the DNA methylome are highly dynamic, whereby exercise leads to hypomethylation, detraining results in hypermethylation, and subsequent retraining can further restore hypomethylation within target loci (13–20). Further, newly acquired myonuclei show promoter DNA hypomethylation within genes related to transcription factor regulation, whereas retained myonuclei exhibit DNA hypomethylation linked to protein turnover (21). Altogether, these studies point to the acute and chronic effects of exercise training on the myonuclei that might alter transcriptional responses and impact cellular metabolism.

Exercise induces lasting mitochondrial adaptations that reshapes cellular energy metabolism. Numerous studies now link energy metabolism to changes in the epigenome, including changes to one carbon metabolism (DNA methylation), alpha-ketoglutarate (DNA demethylation), and acyl-CoA abundance (histone post-translational modifications) (22). In humans, resistance training resulted in a coordinated epigenetic and transcriptional response associated with enhanced cellular bioenergetics (23). Further, recurrent resistance training led to an enhancement of proteins related to aerobic metabolism (24), indicating an indispensable role for mitochondria associated with the muscle memory. In aged mice, endurance training (PoWeR) restored muscle DNA methylation signatures similar to that of young mice, including hypomethylation of mitochondrial electron transport complexes (18, 25). Moreover, newly acquired myonuclei during exercise retraining are distinct from resident myonuclei (21), and have greater hypertrophic capacity and mitochondrial biogenesis (26). Furthermore, exercise training alters the temporal transcriptional and methylation responses, compared to untrained mice (27). This indicates a priming mechanism within the trained muscle resulting in quicker and robust changes in transcription following an acute exercise bout, which are driven by peroxisome proliferator coactivator 1 alpha (PGC-1α)—the master regulator of mitochondrial biogenesis (27). Other candidates linked to the muscle memory point to the myogenic transcription factor MYC that is crucial in myogenesis and energy metabolism (13, 28). Taken together, given the indispensable role of mitochondria as a mediator of the memory of exercise in skeletal muscle, characterization of mitochondrial metabolism might provide insights for ensuring long-term improvements in muscle health and function (22, 29).

The current study aimed to determine how exercise enhances the long-term health of skeletal muscle. We hypothesized that prior endurance training results in a lasting metabolic memory of growth in skeletal muscle, which persists after exercise cessation, and can overcome dietary challenges to support muscle hypertrophy. Utilizing a novel model of voluntary endurance exercise training, detraining, and retraining, we observed that retrained mice exhibited significantly greater muscle mass (12 – 30%) and fiber cross-sectional area (12%) compared to naïve age-matched runners, despite similar exercise exposure. Our research reveals new mechanisms by which endurance exercise shapes the muscle memory to drive growth and energy metabolism. We identified a robust mitochondrial response as a hallmark of the muscle memory regardless of diet and propose that it directly supports and coordinates muscle adaptations with retraining. Thus, our findings provide a detailed view of how the muscle memory is fueled by mitochondrial metabolism, which could be leveraged to combat age-related sarcopenia and metabolic disease.

MATERIALS AND METHODS

Mice and training paradigm

To evaluate the effect of endurance exercise training on skeletal muscle adaptations, eight-week-old male C57BL/6 mice (Charles River Laboratories) were fed ad libitum a control diet (CD; 10% Kcal fat; Research Diets Inc., Cat# D12450H) or a high fat diet (HFD; 45% Kcal fat; Research Diets Inc., Cat# D12451) and placed in cages with either 24-hour access to running wheels (9.5-inch-diameter stainless steel wheel for voluntary wheel running; VWR) or static cages (SED). For the TRAIN cohort, mice were exposed to running wheels (VWR) or static, sedentary cages for 4 weeks (1 cohort; n=13 per group). For the DETRAIN cohort, mice were either endurance-trained (VWR) or sedentary for 4 weeks, followed by a detraining period during which all mice were transferred to static, sedentary cages for 4 weeks (8 weeks total) (1 cohort; n=13 per group). For the RETRAIN cohort, mice were trained (VWR) or sedentary for 4 weeks, followed by detraining for 4 weeks (static cages), and trained (VWR) once again or sedentary for an additional 4 weeks (12 weeks total) (1 cohort; n=13 per group). To isolate the effects of prior exercise, we included a control group that remained sedentary for 8 weeks before undergoing a single 4-week training bout of VWR (naïve VWR, nVWR) (1 cohort; n=9). Body weights were recorded weekly, and body composition (fat and lean mass) was assessed at baseline then every subsequent month for each cohort (EchoMRI-700). To minimize the acute effects of exercise (30), mice were sacrificed using CO₂ inhalation two days (3 cohorts; n=6 per group) after completion of each cohort, and to assess whether these adaptations were a result of a training memory, mice were sacrificed one week (4 cohorts; n=6–7 per group) after completion of each experimental period. After each experimental period, mice were either overnight fasted (n=3 per group) or fed ad libitum (n=3 per group) to evaluate in vivo protein synthesis. Hindlimb leg muscles Tibialis anterior (TA), Extensor Digitorum Longum (EDL), Plantaris (PLANT), Gastrocnemius (GAS), Soleus (SOL), and Levator Ani (LA) were frozen in liquid N₂-cooled isopentane or embedded in OCT and stored at −80°C. All procedures were followed as approved by the Institutional Animal Care and Use Committee at the University of Illinois Urbana-Champaign.

In vivo Protein Synthesis

The levels of muscle protein synthesis were evaluated using the SUnSET method (31, 32), which utilizes the non-radioactive amino acid analog puromycin (Streptomyces alboniger; Millipore Sigma, Cat# P8833). Muscle protein synthesis was measured in the fed and overnight fasted state (3 cohorts; n=3 each) and was given via intraperitoneal injection of 0.040 μmol/g puromycin dissolved in 100μL of PBS 30 min prior to tissue collection according to previously verified studies (31, 32).

Fiber Typing and Cross-sectional Area

The Plantaris (PLANT) muscle (4 independent cohorts, 1 whole cross section per animal, n=5–7 per group) was sectioned at −20°C at a thickness of 10 μm and mounted onto microscope slides. Muscle sections were then allowed to thaw to room temperature (RT) and blocked for 1.5 hours using blocking buffer [5% bovine serum albumin (BSA; Sigma-Aldrich, Cat# A9647) with 0.05% triton X-100 diluted in phosphate buffered saline (PBS)]. Sections were then probed overnight at 4°C with a primary antibody (Ab) cocktail to detect dystrophin (1:200; Abcam Cat# ab15277, RRID:AB_301813), Type 1 fibers (1:50; DSHB Cat# BA-D5, RRID:AB_2235587), Type IIa fibers (1:50; DSHB Cat# SC-71, RRID:AB_2147165), and Type IIb fibers (1:50; DSHB Cat# BF-F3, RRID:AB_2266724) diluted in blocking buffer. Next, sections were washed in PBS and then probed for 1 hour at room temperature with a secondary Ab cocktail containing AlexaFluor633 α-rabbit (1:100; Thermo Fisher Scientific Cat# A-21070, RRID:AB_2535731), AlexaFluor350 α-mouse IgG2b (1:200; Thermo Fisher Scientific Cat# A-21140, RRID:AB_2535777), AlexaFluor568 α-mouse IgG1 (1:200; Thermo Fisher Scientific Cat# A-21124, RRID:AB_2535766), and AlexaFluor488 α-mouse IgM (1:200; Thermo Fisher Scientific Cat# A-10680, RRID:AB_2534062) diluted in blocking buffer. Following final washes, coverslips were added using Clearmount-containing Tris Buffer (Electron Microscopy Sciences, Cat# 17985–12). Whole muscle cross sections were acquired using a Zeiss AxioScan Z.1 Slide Scanner within the Core Facilities at the Carl R. Woese Institute of Genomic Biology. Briefly, images were taken at 10x magnification and stitched together automatically via the slide scanner to achieve whole cross sections of each muscle sample (4 independent cohorts, 1 whole cross section per animal, n=5–7/group). Images were then analyzed using MyoVision (33) and manually confirmed to determine fiber cross sectional area (fCSA) and fiber type distribution.

RNA Extraction, cDNA Synthesis, and qPCR

Total RNA was extracted from the GAS or TA muscles (3 independent cohorts, n=3–6/group) using TRIzol reagent (Thermo-Fisher, Cat# 15596018), ensuring representation of all portions of each muscle. RNA purification was done using the Direct-zol kit (Zymo Research, Cat# R2072) according to the manufacturer’s instructions. Reverse transcription was performed to generate cDNA libraries using the High-capacity cDNA Reverse transcription kit (Applied Biosystems, Cat# 43–688-14). Quantitative PCR was performed using the StepOnePlus Real-Time PCR System with Power SYBR Green PCR Master Mix (Applied Biosystems, Cat# 4368708). Relative DNA concentrations were normalized to housekeeping genes (Rpl7a, Tbp, Actb) and calculated using the ΔΔC_t method. Primer sequences are listed in Suppl. Table 1.

Sequencing (RNA-seq), Mapping, and Analysis

RNA integrity was assessed using a Fragment Analyzer (Advanced Analytical), and samples with RIN>8 were included in the analysis. Polyadenylated mRNA libraries were prepared using the TruSeq Stranded mRNA Sample kit (Illumina) with Unique Dual Indexes to prevent barcode switching. The adaptor-ligated double-stranded cDNAs were amplified for 8 cycles with the Kapa HiFi polymerase (Roche, IA) to reduce the likeliness of multiple identical reads due to preferential amplification. Libraries were quantified using the Qubit (Thermo-Fisher), fragment size was assessed with a Fragment Analyzer, and the barcoded RNAseq libraries were loaded on a NovaSeq lane for cluster formation and sequencing. The libraries were sequenced from one of the fragments for a total of 100bp to a depth of at least 25 million reads per library. The fastq read files were demultiplexed (bcl2fastq v2.20; Illumina, San Diego, CA) and quality checked (FastQC). Alignment and transcript quantification were done using Salmon (v1.10.0) in quasi-mapping mode to the GENCODE vM33 transcriptome (GRCm39) using a decoy-aware indexing method. The transcript-level count was aggregated to gene-level counts using the lengthScaledTPM method from tximport. Gene-level normalization and filtering were conducted using edgeR. Genes with <0.2 counts per million (CPM) in fewer than 3 samples were removed. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, Principal Component Analysis (PCA), and upset plots were obtained using ExpressAnalyst (34). Differentially expressed genes were defined as log2(RPM) value > 0 and Bonferroni adjusted −log10p value >0.7.

Protein Extraction, Immunoblotting, and Citrate Synthase Activity

TA muscle samples were digested using a digestion buffer (50mM Tris-HCl, 250mM Mannitol, 50mM NaCl, 50mM NaF, 1.3mM EDTA, 1mM EGTA, 5mM NaPPi decahydrate, 1mM b-glycerophsophate, 10% w/v glycerol, 1% w/v Tween 20) with a Pierce protease inhibitor (Themo-Fisher, Cat# A32953) and homogenized using a bead beater (MP Biomedicals) (35). Protein concentration was determined using a Bradford assay (BCA; Thermo-Fisher, Cat# 23238). Sample preparation was done using Laemmli buffer (125mM Tris-HCl pH 6.8, 4% SDS, 20% Glycerol, 0.02% bromophenol blue, 10% b-mercaptoethanol) ensuring equal amounts of protein. Muscle lysates (2d post n=3/group; 1wk post n=6–7/group) separated by SDS-PAGE and transferred to polyvinyl difluoride (PVDF) membranes and blocked in a 5% non-fat dry milk (NFDM) tris buffered saline with 0.1% Tween-20 (TTBS) solution for one hour at room temperature. After washing, primary antibody (Ab) for oxidative phosphorylation (OxPhos) (Thermo-Fisher, Cat# 45–8099, RRID:AB_2533835); carnitine palmitoyltransferase 1B (CPT1B) (Cell Signaling Technologies, Cat# 41803, RRID:AB_3698105); hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha and beta (HADHA/B) (Abcam, Cat# ab110302, RRID:AB_10862577); Acyl-CoA dehydrogenase, long chain (ACADL) (Abcam, Cat# ab128566, RRID:AB_11141244); and Puromycin (Sigma-Aldrich, Cat# MABE343, RRID:AB_2566826) were diluted to 1:1000 in a 5% bovine serum albumin (BSA) TTBS solution and membranes were probed overnight at 4°C. Next, membranes were washed and probed with a horseradish peroxidase conjugated secondary [Peroxidase AffiniPure Goat Anti-Mouse IgG (Jackson ImmunoResearch Labs, Cat# 115–035-206, RRID:AB_2338514); Rabbit IgG HRP (Sigma-Aldrich, Cat# GENA934, RRID:AB_2722659)] for one hour at RT. The immunoreactive proteins were detected with enhanced chemiluminescence (Millipore Sigma, Cat# GERPN3004), then images were taken using a ChemiDoc XRS+ with Image Lab Software (BioRad). Protein quantification was done using ImageJ software and normalized to Ponceau S for total protein. Citrate synthase activity was determined from muscle protein lysates using a commercial kit (Millipore Sigma, Cat# CS0720).

Statistical Analysis

All data assumptions (normality, homogeneity of variances, linearity) were met. A linear mixed-effects model was used to assess the effects of exercise, timepoint, and their interaction on body weight and composition (weight, fat and lean mass). Fixed effects included exercise, timepoint, and exercise*timepoint interaction, with subject nested within exercise included as a random effect to account for repeated measures. For within-timepoint post-hoc comparisons, statistical significance was defined as p<0.05 and determined by Bonferroni-corrected pairwise Student’s t-test or Tukey’s HSD, as indicated. Precise p-values are indicated throughout the text and in the figure legends.

RESULTS

Prior endurance training limits weight gain and boosts muscle growth

Muscle memory is thought to promote long-term metabolic benefits associated with recurrent exercise training. To evaluate the effect of a muscle memory we established an exercise training paradigm consisting of VWR TRAINing (4wks), unloading or DETRAINing (+4wks), and VWR-based RETRAINing (+4wks) (Fig. 1A; Suppl. Fig. 1). A separate group of mice underwent VWR without prior training (naïve VWR, nVWR) (Fig. 1A). Using this paradigm, we determined how exercise alters muscle mass 2-days (4, 8, 12 weeks) and how this is retained 1-week after each training cycle (5, 9, 13 weeks), compared to age- and diet-matched controls (Fig. 1A; Suppl. Fig. 1). Following the initial four-week period (TRAIN), sedentary control diet-fed (SED/CD) mice had a significantly increased body weight (BW) (Fig. 1B; Suppl. Fig. 1B, C) and showed a reduction in lean-to-fat mass ratio (Fig. 1C; Suppl. Fig. 1D). Conversely, VWR/CD attenuated weight gain, and while TRAINing reduced lean mass (Fig. 1B; Suppl. Fig. 1B, C), it preserved the lean-to-fat mass ratio (Fig. 1C; Suppl. Fig. 1D). Following the unloading phase (DETRAIN), VWR/CD mice gained weight, primarily coming from fat mass while preserving lean mass, compared to SED/CD (Fig. 1B; Suppl. Fig. 1G-J). However, after exercise RETRAINing utilizing VWR/CD mice showed maintenance of body weight (Fig. 1B), with a significant reduction in fat mass and an increase in lean mass (Fig. 1B, C). Repeated measures ANOVA revealed a significant main effect of timepoint for body weight (F(3, 70.6) = 180.10, p<0.0001), fat mass (F(3, 71.3) = 95.37, p<0.0001), and lean mass (F(3, 70.8) = 64.45, p<0.0001), indicating that all three measures changed significantly over time. A significant main effect of exercise was observed for fat mass only (F(1, 26.2) = 33.09, p<0.0001), while body weight and lean mass were not affected by exercise alone (p = 0.1661 and p = 0.2138, respectively. However, there was a significant exercise*timepoint interaction for both body weight (F(3, 70.6) = 4.08, p<0.0098) and fat mass (F(3, 71.3) = 22.27, p<0.0001), suggesting diverging effects of exercise training over time. No significant interaction was detected for lean mass (p = 0.14). This is relevant, given that the voluntary running distances differed between training periods, where VWR/CD mice ran ~10 Km/day in the initial TRAIN phase and ~6 Km/day during the subsequent RETRAIN phase (Fig. 1D). Furthermore, mice that were endurance trained for 4 weeks without prior exercise (nVWR) show similar reductions in BW and fat mass without the increase in lean mass observed with RETRAINing (Fig. 1B, C), despite having similar running distances (Fig. 1D).

Figure 1. Endurance retraining enhances body composition and muscle growth.

A) Endurance exercise TRAIN-DETRAIN-RETRAIN paradigm. B) Body composition of the RETRAIN (SED/CD and VWR/CD; 1 cohort, n=10/group) and the nVWR cohorts (n=9) indicating stacked fat (yellow) and lean mass (pink), and body weight (lines, BW) changes across each cycle. C) Relative body composition (% BW) of the RETRAIN (SED/CD and VWR/CD; 1 cohort, n=10/group) and the nVWR cohorts (n=9) indicating fat (yellow) and lean proportions (pink) and BW changes. Two-way ANOVA (time*exercise) showing within time point comparison of means for fat and lean mass, pairwise Student’s t-test for 0-, 4-, and 8- weeks and Tukey’s HSD for 12-weeks. D) Daily average running distances (Km/day) by week (left) and average running distance (Km/day) (right) during each endurance training phase from the 2 cohorts (n=9–10). E) Mouse hindlimb muscle diagram and relative muscle weights (ratio to BW) of the TA muscle. F–I) Relative muscle weights (ratio to BW) across all cohorts for hindlimb muscle including E) TA, F) EDL, G) GAS, H) PLANT, and I) SOL. J–N) Absolute muscle weights of J) TA, K) EDL, L) GAS, M) PLANT, and N) SOL with effect size estimation between RETRAIN and TRAIN cohorts (4 independent cohorts shown, n=6–7/group). Data are presented as means ± SD. Within time point comparison of means using pairwise Student’s t-test for 5, and 9 weeks, and Tukey’s HSD for 13 weeks.

Next, we determined how the relative weight (Fig. 1E–I; Suppl. Fig. 1E, F, K, L) of hindlimb muscles—including Tibialis Anterior (TA), Extensor Digitorum Longus (EDL), Gastrocnemius (GAS), Plantaris (PLANT), and Soleus (SOL)—adapt to prior and recurrent exercise training. Endurance exercise (VWR/CD) TRAINing increased the relative weight of the soleus (SOL) (Fig. 1I) but no other hindlimb muscles (Fig. 1E–H; Suppl. Fig. 1E), indicating a maintained contribution to body weight. After 4 weeks of unloading (DETRAIN), while the contribution of muscle weight to total body weight was reduced compared to TRAINing, relative to age-matched SED/CD the relative muscle weights of TA, EDL, GAS, PLANT, and SOL remained higher in VWR/CD mice (Fig. 1E–I; Suppl. Fig. 1K) suggesting a preservation of the relative contribution of body weight for each muscle in mice that completed exercise TRAINing. Despite lower daily average activity than the initial TRAINing phase (Fig.1D), prior VWR exercise potentiated muscle regrowth with RETRAINing, as demonstrated by greater relative weights of TA, GAS, PLANT, and SOL (Fig. 1E–I). Endurance exercise without prior training (nVWR) only increased relative muscle weights of the GAS and PLANT (Fig. 1G, H), further highlighting the role of prior exercise at inducing a memory of exercise in skeletal muscle.

We then determined how the absolute weight (Fig. 1J–N) of hindlimb muscles adapted to prior and recurrent exercise training. After TRAINing, only the TA muscle weight decreased in VWR/CD mice (Fig. 1J; Suppl. Fig. 1F). With DETRAINing, absolute muscle weights did not differ between groups (Fig. 1J–N, Suppl. Fig. 1L). However, after RETRAINing, the absolute weight of TA, GAS, PLANT, and SOL, but not EDL, were significantly higher in VWR/CD mice, compared to age-matched SED/CD (Fig. 1J–N). However, nVWR did not result in raw muscle weight increases, instead it reduced SOL mass, compared to VWR/CD (Fig. 1J–N). This is significant given that neither the relative nor absolute weight of non-hindlimb muscles, such as the Levator Ani (LA) muscle, which has a high abundance of anabolic hormone receptors, was changed by endurance exercise across all training cycles (Suppl. Fig. 1M, N). Further, a comparison of the anabolic responses between either training cycles (RETRAIN vs. TRAIN; nVWR vs. VWR) confirmed the dramatic enhancement of muscle growth in the TA (Fig. 1J), EDL (Fig. 1K), GAS (Fig. 1L) and PLANT muscles (Fig. 1M).

Endurance training and retraining alter fiber distribution and hypertrophy

To determine whether muscle growth following endurance training cycles results from increased fiber cross-sectional area (fCSA), we analyzed whole Plantaris muscle fiber size and distribution using immunofluorescence (IF) (Fig. 2). This approach allowed us to assess adaptations after initial TRAINing, how these changes persist after DETRAINing, and whether prior training primes the muscle for enhanced RETRAINing responses. Representative images for SED/CD, VWR/CD, and nVWR/CD mice after each endurance training cycle are shown in Figure 2A. Endurance TRAINing led to a modest shift in fiber size, resulting in a reduced frequency of fibers between 250–750 μm² while increasing the frequency of fibers larger than 1000 μm² (Suppl. Fig. 2A), while these trends were not significant it does shed light on the hypertrophic stimulus of exercise TRAINing. However, the overall fiber type distribution largely overlapped with that of SED/CD, and we observed no changes in relative fiber distributions (Fig. 2B–D). Despite this, TRAINing induced hypertrophy in oxidative type IIa (Fig. 2B) and intermediate type IIx fibers (Fig. 2C). This suggests that endurance exercise promotes selective fiber growth without significantly altering global fiber type distribution.

Figure 2. Endurance retraining enhances muscle fiber composition and hypertrophy.

A) Representative cross-sections of whole PLANT muscle from SED/CD (top) and VWR/CD (bottom) groups across TRAIN (5wks; left), DETRAIN (13wks; middle), and RETRAIN/nVWR cohorts (13wks; right). B–D) Distribution (left) and fiber Cross-Sectional Area (fCSA) (right) for B) type IIa, C) IIx, and D) IIb muscle fibers in PLANT (4 independent cohorts, 1 whole cross section per animal, n=5–7/group) across cohorts. Data are presented as means ± SD. Within time point comparison of means using pairwise Student’s t-test for 5, and 9 weeks, and Tukey’s HSD for 13 weeks.

After 4 weeks of inactivity, fiber size distribution returned to a pattern similar to that of SED/CD (Fig. 2B–D). DETRAINing also reduced the percentage of type IIx fibers (Fig. 2C) but did not significantly affect the proportions or size of type IIa or type IIb fibers (Fig. 2B, D). These findings suggest that while endurance TRAINing initiates adaptations, fiber hypertrophy, and distribution shifts, these may not be maintained without continued exercise. However, prior TRAINing appeared to prime the muscle for enhanced RETRAINing responses. RETRAINing led to an increase in the proportion of oxidative IIa fibers (Fig. 2B) while decreasing glycolytic IIb fibers (Fig. 2D), indicating a shift toward a more oxidative muscle phenotype. Additionally, RETRAINing significantly increased the fCSA of both oxidative type IIa (Fig. 2B) and intermediate type IIx fibers (Fig. 2C). While there appears to be a numerical increase of larger fibers (>2250 μm²) in RETRAINED mice compared to sedentary controls (who had a greater proportion of fibers <1750 μm²), these shifts were not significant (Suppl. Fig. 2A). However, these trends further supports the idea that prior training enhances muscle growth with repeated exercise (Suppl. Fig. 2A). These findings highlight that while endurance TRAINing alone does not drastically alter fiber distribution, it may prime the muscle for enhanced energy production, thus leading to greater adaptation upon RETRAINing, resulting in larger and more oxidative fibers. Thus, despite lower voluntary endurance exercise volumes during the RETRAIN phase, this period was associated with an absolute increase in hindleg muscle mass and fiber type-specific hypertrophy that may be driven by enhanced mitochondrial energy production. This is further supported by the lack of changes to type IIa and IIx fCSA in nVWR mice, compared to age-matched SED/CD (Fig. 2A–D). As prior endurance training was essential for potentiating retraining adaptations (Fig. 1 & 2), the nVWR group was excluded from further analysis, given its limited relevance to muscle memory.

To determine if these changes in muscle growth were driven by increased muscle protein synthesis, we evaluated the incorporation of puromycin (31, 32) in skeletal muscle (TA) following TRAIN-DETRAIN-RETRAIN under both fasted and fed conditions (Suppl. Fig. 2B, C). Representative blots are shown for each cohort (Suppl. Fig. 2B) and analysis reveals that endurance RETRAINing does not impact muscle protein synthesis as puromycin incorporation is largely affected by the fasted/fed status of these mice regardless of training status (Suppl. Fig. 2C). Additionally, expression of protein degradation genes was not altered (Suppl. Fig. 2D), suggesting the lasting adaptations may not be the result of reduced protein degradation.

Prior exercise primes skeletal muscle to increase mitochondrial metabolism with recurrent exercise.

To understand how exercise RETRAINing shapes muscle cellular coordination, we aimed to characterize the transcriptional memory of endurance exercise following TRAINing (4wks), DETRAINing (8wks), and RETRAINing (12wks) (Fig. 3). Recent data from the Molecular Transducers of Physical Activity Consortium (MoTrPAC) (30) has shown that the acute effects of exercise persist for up to 48hrs. Thus, we characterized the gene expression of GAS muscle 48 hrs-post exercise using bulk RNA sequencing (RNA-seq). Global gene expression profiles (differentially expressed genes; DEGs) of muscle samples from VWR/CD and SED/CD controls (assessed with principal component analysis; PCA) (Fig. 3A) demonstrate the effects of the initial TRAIN (circles) and final RETRAIN phases (squares), with both phases showing differential signatures after each exercise training bout (Fig. 3A), and unique DEGs (Fig. 3B; Suppl. Fig. 3A-C). Additionally, during the DETRAIN phase, muscle transcription shows an expression pattern more closely associated with untrained signatures (Fig. 3A), but distinct from SED/CD age-matched controls (Fig. 3B; Suppl. Fig. 3B). While a greater number of DEGs were identified after DETRAIN (~4700 genes) (Fig. 3B), RETRAIN increases the expression of ~330 unique genes (Fig. 3B). We also observed an overlap across all endurance training cohorts with 36 upregulated genes and 1 downregulated gene (Fig. 3B; Suppl. Fig. 3C), most of them associated with tyrosine-kinase signaling (Efnb2, Lnx1, Itpr2, Plcb4, Plcl1, Sh2d3c), intracellular protein trafficking (Arhgef3, Plekhg1, Rab27), and myogenic processes (Myh1, Notch1, Osmr, S1p1r, Tanc1, Klf6). Thus, endurance training cycles elicit consistent but unique effects on gene transcription, which may underlie the transcriptomic memory.

Figure 3. Endurance exercise retraining optimizes mitochondrial metabolism.

Differential gene expression analysis of GAS muscle (3 independent cohorts, 2d-post exercise, n=3/group) between VWR/CD and SED/CD showing A) Principal component analysis (3D PCA) of DEGs across TRAIN (circles), DETRAIN (diamonds), and RETRAIN cohorts (squares) and B) upset plot. DEGs adjusted p< 0.2. C) Venn diagram of KEGG pathways across all cohorts, compared to respective SED/CD. D) Circular plot showing the submitochondrial mapping of RETRAIN DEGs to MitoCarta 3.0. Log2 fold change vs. SED/CD. Inter Membrane Space: IMS; Mitochondrial Inner Membrane: MIM; Mitochondrial Outer Membrane: MOM. E) Representative immunoblots and F) quantification of OxPhos complexes, CPT1B, HADHA/B, and ACADL proteins in TA muscle (3 independent cohorts, 1wk-post exercise, n=6–7/group). G) Citrate synthase activity across in TA across all cohorts (3 independent cohorts, 1wk-post exercise, n=6–7/group). Data are presented as means ± SD. Within time point comparison of means using pairwise Student’s t-test.

Among the pathways associated with TRAIN (vs. age-matched SED/CD) using the Kyoto Encyclopedia of Genes and Genomes (KEGG), we identified the NOD-like and NOTCH signaling related to inflammation, maintenance, and repair of muscle cells (Fig. 3C; Suppl. Fig. 3A). Moreover, muscle unloading (DETRAIN) results in alterations to numerous pathways related to lipid metabolism, cell growth, and inflammation (Fig. 3C; Suppl. Fig. 3B). However, the number of DEGs was smaller after endurance RETRAINing compared to DETRAIN (Fig. 3B), the former resulted in dramatic upregulation of mitochondrial genes (Fig. 3C; Suppl. Fig. 3C) including oxidative phosphorylation (OxPhos) associated genes (Fig. 3C), with little overlap in pathways across all training cycles (Fig. 3C). Expression of Pgc-1α in muscle a week removed was not different from age-matched controls (Suppl. Fig. 3D). DNA methylation modifier Dnmt1 was reduced after RETRAINing (Suppl. Fig. 3E), but no changes were observed in demethylases Tet1, 2, or 3 (Suppl. Fig. 3E).

Categorization of the sub-mitochondrial localization of DEGs upregulated after endurance RETRAINing was done utilizing MitoCarta (36) (Fig. 3D). Our results show that ~94% (149 DEGs) of the upregulated mitochondrial genes correspond to Mitochondrial Inner Membrane (MIM) Matrix, Intermembrane space (IMS), and Mitochondrial Outer Membrane (MOM) proteins (Fig. 3D). Thus, our data shows that prior training is important to enhance mitochondrial gene transcription with retraining.

Given the robust enhancement of mitochondrial genes following retraining, we investigated how these adaptations remained a week after each exercise cycle (5, 9, 13wks) by assessing mitochondrial protein levels (Fig. 3E, F). One week after the initial TRAINing period (5wk), OxPhos complex CV was significantly reduced relative to age-matched SED/CD (Fig. 3E, F). However, following DETRAINing (9wk), OxPhos CII was significantly elevated relative to age-matched SED/CD (Fig. 3E, F). Following RETRAINing (13wk), OxPhos complex CIV remained elevated, indicating a sustained enhancement of mitochondrial respiratory capacity (Fig. 3E, F). To determine how endurance training cycles impacted mitochondrial fat oxidation (FAO), we investigated the protein levels of Carnitine-Palmitoyl Transferase B (CPT1B), a rate-limiting step in FAO (37), Hydroxyacyl-CoA Dehydrogenase Trifunctional Multienzyme Complex Subunit Alpha/Beta (HADHA/B), and the Long-Chain Acyl-CoA Dehydrogenase (ACADL) (Fig. 3E, F). Muscle CPT1B was reduced after DETRAINing but non-significantly upregulated a week removed from RETRAINing (Fig. 3E, F). Similarly, HADHA/B, which catalyzes the final steps of FAO (38), shows no differences across time, with only a trend to increase after endurance RETRAINing (Fig. 3E, F). Muscle ACADL, a marker of fatty acid utilization (39), was not altered at TRAINing or DETRAINing but was significantly increased following RETRAINing (13wk) relative to age- and diet-matched SED/CD (Fig. 3E, F).

To determine the overall activity of the tricarboxylic acid cycle (TCA), we assessed citrate synthase (CS) activity—a useful standard for mitochondrial density and oxidative capacity (40). Our data shows that endurance RETRAINing significantly increased mitochondrial activity (Fig. 3G), reinforcing the notion that prior endurance exercise primes skeletal muscle for efficient oxidative metabolism.

Prior endurance training negates high fat-diet feeding to enhance muscle growth.

Our data as well as that of others (11, 14, 15) demonstrate that prior exercise training improves body composition and enhances muscle growth. However, how the muscle memory can persist and combat the effects of an obesogenic diet has not been assessed. Using our endurance training paradigm (Fig. 1), we evaluated the effects of overnutrition by feeding mice a HFD (45% kcal fat) (Fig. 4A; Suppl. Fig. 4). After the initial TRAIN period, much like the CD-fed mice (Fig. 1), endurance-trained HFD-fed mice (VWR/HFD) show an attenuation of weight and fat accumulation, compared to SED/HFD (Fig. 4B; Suppl. Fig. 4A-D). However, unlike the VWR/CD mice, VWR/HFD mice do not experience a reduction in lean mass after TRAINing (Fig. 4B; Suppl. Fig. 4B, D). After the DETRAINing (8wks), VWR/HFD mice show greater fat mass accumulation (Fig. 4B; Suppl. Fig. 4G-J), but maintain a significantly lower BW, compared to age- and diet-matched SED/HFD (Fig. 4B; Suppl. Fig. 4H-J). After RETRAINing (12 weeks), SED/HFD mice show a progressive increase in fat-to-lean ratio with ~40% of BW coming from fat mass (Fig. 4C). Conversely, while endurance RETRAINing did not appear to reduce BW (Fig. 4B), VWR/HFD mice show increased lean mass and significantly lower fat-to-lean ratio with ~20% of BW coming from fat mass (Fig. 4B, C). Repeated measures ANOVA showed a significant main effect of exercise on body weight (F(1, 18.0) = 14.67, p= 0.0012) and fat mass (F(1, 18.0) = 17.43, p= 0.0006), but not on lean mass (p= 0.7998). A significant main effect of timepoint was observed for all three outcomes: body weight (F(3, 54.0) = 349.97, p< 0.0001), fat mass (F(3, 54.0) = 153.42, p< 0.0001), and lean mass (F(1, 54.0) = 207.67, p< 0.0001), indicating changes over time. Significant exercise*timepoint interactions were also detected for body weight (F(3, 54.0) = 20.85, p< 0.0001) and fat mass (F(1, 54.0) = 17.02, p< 0.0001), suggesting that the effects of exercise varied across timepoints. No significant interaction was observed for lean mass (p= 0.3143). Further, similarly to CD-fed mice, the running distances differed between training periods, where VWR/HFD mice ran ~8 Km/day in the TRAIN phase and ~5 Km/day during the RETRAIN phase (Fig. 4D). Thus, while VWR/HFD mice had lower voluntary exercise volumes in the RETRAIN phase, this period was associated with an absolute increase in lean mass and weight maintenance.

Figure 4. Endurance retraining negates an obesogenic diet to promote muscle growth.

A) Mice on an obesogenic HFD were assigned to endurance exercise protocols of TRAIN-DETRAIN-RETRAIN or sedentary conditions. B) Body composition of the RETRAIN HFD cohort indicating stacked fat (yellow) and lean mass (pink), and body weight (lines, BW) changes across each phase. C) Relative body composition (% BW) of the RETRAIN cohort indicating fat (yellow) and lean proportions (pink) and BW changes. Inset above indicates a two-way ANOVA (time*exercise) showing within time point comparison of means for fat and lean mass, pairwise Student’s t-test for 0-, 4-, 8-, and 12-weeks with indicated p-values. D) Daily average running distances (Km/day) by week (left) and average running distance (Km/day) (right) during each endurance training phase from 1 cohort (n=10). E–I) Relative muscle weights (ratio to BW) across all cohorts for hindlimb muscles, including E) TA, F) EDL, G) GAS, H) PLANT, and I) SOL. J–N) Absolute muscle weights of J) TA, K) EDL, L) GAS, M) PLANT, and N) SOL with effect size estimation between RETRAIN and TRAIN cohorts (3 independent cohorts, n=6–7/group). Data are presented as means ± SD. Within time point comparison of means using pairwise Student’s t-test.

Next, we examined how the significant attenuation of fat accumulation with endurance training (VWR/HFD) is associated with hindlimb muscle growth, compared to age- and diet-matched SED/HFD (Fig. 4E–I; Suppl. Fig. 4E-F, K-L). Our data shows that the relative weight of TA (Fig. 4E), EDL (Fig. 4F), GAS (Fig. 4G), PLANT (Fig. 4H), and SOL (Fig. 4I) muscles (Suppl. Fig. 4E-F, K-L) are increased in VWR/HFD mice across all timepoints, compared to SED/HFD controls. Further, although obesogenic conditions attenuated the effects of prior training, we demonstrate that endurance RETRAINing enhances muscle growth even under obesogenic HFD (Fig. 4J–N). Absolute masses of the TA (Fig. 4J), GAS (Fig. 4L), PLANT (Fig. 4M), and SOL (Fig. 4N) are increased compared to SED/HFD. The effect sizes of RETRAIN vs. TRAIN periods showed greater hypertrophy for only the GAS, which is highly recruited during VWR (41), and the SOL (Fig. 4L, N)—a largely oxidative muscle (42). No changes were observed in the LA muscle across cohorts (Suppl. Fig. 4M, N). Thus, prior exercise primes the muscle (GAS and SOL) for subsequent exercise, and despite reduced total exercise volume (Fig. 4D), both body weight and fat mass were below that of age- and diet-matched sedentary controls (Fig. 4B, C).

RETRAINing enhances muscle oxidative fiber size and composition despite obesogenic diet feeding.

Endurance RETRAINing preserves body weight and increases lean mass in HFD-fed mice. To determine how an obesogenic diet shapes muscle fiber hypertrophy, we imaged whole PLANT muscle cross sections from VWR/HFD and SED/HFD mice (Fig. 5A). While no significant shifts in global fiber size distribution were observed following TRAIN and DETRAIN, we noted a marginal increase in large fibers (<3000 μm²) in VWR/HFD, compared to age- and diet-matched SED/HFD (Suppl. Fig. 5A). Specifically, VWR/HFD mice show a shift in PLANT composition and size favoring oxidative type IIa (Fig. 5B), with no changes to IIx fiber distribution but an increase in their size (Fig. 5C), and a reduction in fast-twitch type IIb fibers while marginally increasing their size (Fig. 5D), compared to age- and diet-matched SED/HFD. This suggests that the increased availability of dietary fat is utilized more during TRAINing to produce energy for exercising muscle. Further, DETRAINing reverted fiber distribution of all fiber types in VWR/HFD mice similar to SED/HFD controls (Fig. 5B–D), but type IIx and a trend for IIa fCSA increase was observed, compared to age- and diet-matched SED/HFD (Fig. 5B, C). Lastly, subsequent exercise RETRAINing increases the distribution and fCSA of oxidative type IIa (Fig. 5B) and intermediate type IIx fibers (Fig. 5C) while reducing the distribution and enhancing the size of type IIb fiber size in VWR/HFD mice (Fig. 5D).

Figure 5. Endurance training enhances oxidative fibers while retraining optimizes oxidative muscle fibers under obesogenic conditions.

A) Representative cross-sections of whole PLANT muscle from SED/HFD (top) and VWR/HFD (bottom) groups across TRAIN (left), DETRAIN (middle), and RETRAIN cohorts (right). B–D) Distribution (left) and fiber Cross-Sectional Area (fCSA) (right) for B) type IIa, C) IIx, and D) IIb muscle fibers in PLANT (3 independent cohorts, 1 whole cross section per animal, n=6–7/group) across cohorts. Data are presented as means ± SD. Within time point comparison of means using pairwise Student’s t-test.

Protein accretion assessed by puromycin incorporation (31, 32) into the muscle (TA) shows no differences with TRAINing and a non-significant reduction with DETRAINing compared to fasted age- and diet-matched SED/HFD controls (Suppl. Fig. 5B, C). Importantly, endurance RETRAINing enhances muscle protein accretion in both the fasted and fed state relative to the age- and diet-matched SED/HFD controls (Suppl. Fig. 5B, C). Further, no expression changes were observed in protein degradation genes compared to age- and diet-matched sedentary controls (Suppl. Fig. 5D).

Prior exercise negates high fat-diet to prime skeletal muscle mitochondrial gene expression with recurrent training

While our data (Fig. 3) and others have demonstrated how exercise shapes a transcriptional memory in skeletal muscle, how this memory persists during an obesogenic diet challenge has not been determined. To do this, we assessed the muscle transcriptional memory in endurance-trained HFD-fed mice (Fig. 6; Suppl. Fig. 6). Like mice fed a CD, global DEG profiles of muscle assessed with PCA demonstrate the effects of the initial TRAIN (circles) and final RETRAIN phases (squares), with both phases showing differential signatures after each exercise training bout (Fig. 6A) and unique DEGs (Fig. 6B; Suppl. Fig. 6A-I). Additionally, during the DETRAIN phase, the muscle transcriptome shows a pattern more closely associated with untrained SED/HFD (Fig. 6A), resembling VWR/CD mice (Fig. 3A). Unlike the VWR/CD mice, which had the majority of DEGs following DETRAINing (Fig. 3B), VWR/HFD had a reduced number of DEGs during this period (Fig. 6B). Moreover, a greater number of unique DEGs were identified after TRAIN (~2400 genes) and RETRAIN periods (~1198 genes) (Fig. 6B; Suppl. Fig. 6B-C, H-I). We also observed a minor overlap across all endurance training cohorts with 16 upregulated genes and 3 downregulated genes (Fig. 6B), most of which associated with AKT/PKB signaling and intracellular protein trafficking (Dock6, Dock9, Plekhg1, Prex2, Rasgrp2, Stxbp6), Notch signaling and Proteostasis (Dll4, Trim12a, Trim25), as well as phospholipid (Abcb1a) and oligopeptide transport (Slc15a5). Thus, endurance training cycles negate HFD feeding to produce consistent and unique effects on muscle gene expression that underlies the transcriptomic memory.

Figure 6. Muscle transcriptional memory persists during an obesogenic diet challenge.

Differential gene expression analysis of GAS muscle (3 independent cohorts, 2d-post exercise, n=3/group) between VWR/HFD and SED/HFD showing A) Principal component analysis (3D PCA) of DEGs across TRAIN (circles), DETRAIN (diamonds), and RETRAIN cohorts (squares) and B) upset plot. DEGs adjusted p< 0.2. C) Venn diagram of KEGG pathways across all cohorts, compared to respective SED/HFD. D) Circular plot showing submitochondrial mapping of RETRAIN DEGs to MitoCarta 3.0. Log2 fold change vs. SED/CD. Inter Membrane Space: IMS; Mitochondrial Inner Membrane: MIM; Mitochondrial Outer Membrane: MOM. E) Representative immunoblots and F) quantification of OxPhos complexes, CPT1B, HADHA/B, and ACADL proteins in TA muscle (3 independent cohorts, 1wk-post exercise, n=6–7/group). G) Citrate synthase activity across in TA across all cohorts (3 independent cohorts, 1wk-post exercise, n=6–7/group). Data are presented as means ± SD. Within time point comparison of means using pairwise Student’s t-test.

Among the KEGG pathways associated with VWR/HFD after TRAINing (vs. age-matched SED/HFD), we identified the amino acid, glycerolipid, and glycerophospholipid metabolism, transcriptional regulation, and cell contraction are enhanced relative to SED/HFD (Fig. 6C; Suppl. Fig. 6A-C). Moreover, muscle unloading (DETRAIN) resulted in alterations to numerous pathways related to extracellular matrix-receptor interactions, PI3K-AKT signaling, and fatty acid metabolism (Fig. 6C; Suppl. Fig. 6D-F). Interestingly, similarly to the effects of CD-fed endurance RETRAIN (VWR/CD) (Fig. 3C; Suppl. Fig. 6G-I), HFD-fed endurance RETRAINing (VWR/HFD) resulted in a significant upregulation of mitochondrial genes including OxPhos, with minor overlap in pathways across all training cycles (Fig. 6C; Suppl. Fig. 6H-I). Expression of the mitochondrial biogenesis factor Pgc-1α was elevated following both TRAINing and RETRAINing (Suppl. Fig. 6J), suggesting that an obesogenic diet challenge may enhance mitochondrial priming. Further, no major changes in DNA methylation modifiers were observed, with the exception of Tet3 that was reduced after RETRAINing (Suppl. Fig. 6K).

Our study revealed diet-specific changes to the muscle transcriptome (Suppl. Fig. 7A-C) with a robust enhancement of mitochondrial gene expression following RETRAINing (Fig. 3C; 6C; Suppl. Fig. 7C-D). Next, we categorized the sub-mitochondrial localization of DEGs upregulated after endurance RETRAINing using Mitocarta (36) (Fig. 6D). We show that ~93% (299 DEGs) of mitochondrial genes are upregulated corresponding to MIM Matrix, IMS, and MOM proteins (Fig. 6D). Thus, similar to VWR/CD mice (Fig. 3), our data shows that prior endurance training overrides the deleterious effects of HFD to enhance mitochondrial gene transcription with a recurrent training cycle.

Given most pathways enhanced by RETRAINing are related to energy production, we investigated the effect of exercise with concurrent HFD feeding on the levels of OxPhos and FAO proteins (Fig. 6E, F). A week following exercise TRAINing, the relative expression of OxPhos CV was reduced (Fig. 6E, F). After a period of DETRAINing, OxPhos CII was elevated relative to age- and diet-matched SED/HFD (Fig. 6E, F). Interestingly, a week following RETRAINing resulted in a reduction in the expression of OxPhos CI relative to age- and diet-matched SED/HFD (Fig. 6E, F).

Measurement of skeletal muscle FAO proteins CPT1B and HADHA/B showed reduced levels a week removed from TRAINing, and following DETRAINing, these FAO proteins were increased (Fig. 6E, F). Following RETRAINing, the levels of CPT1B and HADHA/B showed no differences compared to age- and diet-matched sedentary controls (Fig. 6E, F). The levels of ACADL were reduced a week removed from TRAINing compared to age- and diet-matched controls (Fig. 6E, F). Importantly, ACADL was increased following DETRAINing and remained elevated with RETRAINing (Fig. 6E, F). Notably, despite the robust transcriptional enhancement of OxPhos genes, CS activity did not increase following RETRAINing in VWR/HFD mice (Fig. 6G). This contrasts with the control diet condition (Fig. 3G), where RETRAINing elevated CS activity alongside muscle growth. This may indicate that while transcriptional control of mitochondrial genes is rewired by exercise RETRAINing, not all proteins are required to support energy demands. Conversely, we show that, under obesogenic conditions, fatty acid utilization may be favored during DETRAINing and RETRAINing to support energy production.

DISCUSSION

What began as the idea of completing a learned task with little mental input, muscle memory has evolved to encompass the anabolic and metabolic adaptations in skeletal muscle. We aimed to determine how prior endurance training elicits a lasting memory that primes enhanced adaptations in skeletal muscle upon endurance retraining. Here, we utilized novel endurance training, detraining, and retraining cycles to assess the memory of exercise in hindlimb muscles. Our studies show that prior endurance training imparts a lasting memory that optimizes mitochondrial metabolism. TRAINing attenuated weight gain caused by sedentarism and negated an HFD challenge to increase muscle fCSA. Although myofiber size and distribution adaptations were not retained, the initial TRAINing phase led to increased relative muscle mass throughout the DETRAINing period. Further, despite reduced running volume during RETRAINing, we uncovered a robust enhancement of mitochondrial metabolism (gene and protein expression), suggesting an optimization of mitochondrial energy production, which led to increased muscle growth with subsequent exercise. These findings highlight the potential of exercise memory as a powerful mechanism to boost metabolic resilience and improve responses to future exercise interventions.

Prior studies have demonstrated that the muscle memory encodes the molecular control of muscle growth or anabolism with recurrent exercise training. In humans, resistance exercise reloading following 7-week loading and 7-week unloading periods is associated with a greater increase in lean mass (14). Utilizing a low-intensity weighted wheel running in mice increased fCSA of type I and IIa fibers in the oxidative SOL muscle, but reduced the proportion of IIa fibers in favor of the more oxidative type I fibers (11). Moreover, using a similar training paradigm, evidence points to enhanced growth of the PLANT muscle with recurrent training, increasing type IIa fiber size but failing to alter the relative proportion of this more oxidative fiber in the mixed GAS (15). In the current study, we utilized larger running wheels compared to prior studies (9.5 vs. 4 in. (9, 11, 15, 16, 21, 25, 43)) to demonstrate how voluntary endurance training can shape the growth of hindlimb muscles. We present definitive evidence that exercise retraining, regardless of diet, results in greater muscle growth than the initial training phase, despite a lower endurance training volume during the retraining phase. Further, we show that endurance retraining results in a glycolytic-to-oxidative shift of the PLANT muscle with oxidative type IIa and intermediate IIx fibers size growth regardless of diet, suggesting that prior endurance training primes the muscle to recover and adapt to subsequent retraining. This is further supported by our finding in nVWR mice that received no prior 4-week training and did not recapitulate the muscle growth observed with retraining. Including nVWR mice as an age-matched comparator provides a rigorous test of exercise-induced muscle memory, rather than 8-week continuous training, as it isolates the effect of prior training history. Remarkably, retraining-induced adaptations persisted even one week after exercise cessation, underscoring the lasting effect of prior exercise in priming skeletal muscle for enhanced muscle growth. Chronic consumption of obesogenic diets (like HFD) has been associated with reduced muscle protein synthesis and fCSA, contributing to muscle atrophy and decline. HFD also produces a fiber-type shift toward glycolytic, less oxidative fibers and reduces fiber size across all muscle fibers, further diminishing endurance capacity and metabolic flexibility (44–46). However, limited evidence exists exploring the effects of HFD on the muscle memory (47). Our findings not only enhance the prior evidence that recurrent endurance exercise potentiated muscle growth but also expands our current understanding by showing that this adaptive hypertrophy persists even under an obesogenic challenge—this highlights the muscle’s primed capacity for recovery and growth with retraining (48–50).

Among the proposed mechanisms regulating muscle memory, research has shown that PGC-1α mediates mitochondrial remodeling and epigenetic responses following acute exercise in previously exercise trained mice (27). The robust and swift expression of Pgc-1α in the trained muscle appears to support the rewiring of muscle bioenergetic coordination, supporting the muscle memory. Optimal fuel utilization in the mitochondria with exercise generates metabolites and proteins (22) capable of altering the epigenome, but ablation of muscle PGC-1α negates the effects of training on DNA methylation and the acute transcriptional responses to exercise (27). Further, exercise has been reported to reduce DNA methylation of Pgc-1α, mitochondrial transcription factor A (TFAM), and citrate synthase (CS) promotor regions (51). Our data extends the muscle memory beyond transcriptional and epigenetic regulation by highlighting the optimization of mitochondrial metabolism to enhance muscle growth. We show that muscle weight continued to increase with RETRAINing, suggesting that enhanced mitochondrial gene expression alone may be sufficient to drive growth without a concurrent rise in mitochondrial TCA (CS) activity.

Additionally, we showed that RETRAINing resulted in a glycolytic-to-oxidative shift of the PLANT muscle with the more oxidative type IIa and intermediate IIx fibers increasing in cross-sectional area, suggesting that prior endurance TRAINing primed the muscle to recover and adapt to subsequent RETRAINing. A striking finding is that, regardless of diet, there is a robust upregulation of mitochondrial genes and pathways following RETRAINing. Similarly, Lee and colleagues (26) utilized a model of resistance training in rats and reported an oxidative phenotype and greater mitochondrial density following resistance retraining 2-days post exercise. Further, this study reported enhanced mitochondrial succinate dehydrogenase (SDH) and Cytochrome C oxidase (COX) activity following the resistance retraining, suggesting the enhanced growth may be directly tied to mitochondrial energy production. Previous studies have identified transcriptomic shifts in HFD-fed muscles including downregulations of genes involved in oxidative metabolism, myogenesis, and structural maintenance, coupled with the upregulation of inflammatory and proteolytic pathways (52–54). Our data shows endurance RETRAINed mice fed a normal or obesogenic diet had a large increase of genes related to OxPhos and mitoribosomal function, highlighting the key role of mitochondria in supporting the exercise memory in skeletal muscle. Mapping of upregulated genes with MitoCarta (36) revealed the submitochondrial compartments altered by RETRAINing, supporting the idea that the mitochondria serve a central role in encoding the memory of exercise. Similarly, Furrer and colleagues (27) suggest that mitochondrial biogenesis and metabolic flexibility persist after detraining and facilitate metabolic adaptation upon retraining.

During exercise, muscle fatty acid transport proteins in the plasma membrane increase by ~75%, with the majority of fatty acids providing energetic support to the working muscle (55). In individuals who are obese, reduced muscle lipid oxidative capacity has been linked to reduced metabolic flexibility of the muscle and impaired insulin sensitivity. At the mitochondrial level, HFD alters mitochondrial density, dysregulates oxidative capacity, and alters lipid handling, leading to the accumulation of intramyocellular lipids and increased reactive oxygen species production (56–59). These adaptations hinder the ability of the muscle to respond to anabolic stimuli, making strategies that fine-tune oxidative capacity highly relevant to counter the deleterious effects of HFD in muscle plasticity. Our studies found that fatty acid metabolism is central to the muscle memory, as shown by increased ACADL levels after RETRAINing, even under HFD conditions. While we do see a robust enhancement of gene expression related to OxPhos regardless of diet, citrate synthase activity increased only under control diet, suggesting fatty acid oxidation may compensate for metabolic stress. This is important given that data suggests individuals with obesity are unable to increase their ability to utilize fatty acids during HFD feeding, compared to lean counterparts, suggesting an increase in overall metabolic health (60). Our results align with the metabolic improvements by supporting enhanced long-chain fatty acid oxidation capacity (ACADL) with recurrent exercise training. These findings highlight the complexity of skeletal muscle plasticity, where RETRAINing induced muscle hypertrophy despite differences in mitochondrial enzyme activity, demonstrating that prior endurance training primes muscle for growth even in the presence of altered metabolic demands. Although we did not directly measure mitochondrial respiration or density, the observed molecular and enzymatic changes provide strong indirect evidence for enhanced oxidative capacity with retraining. Future studies should explore how manipulating specific metabolic pathways during RETRAINing can further optimize muscle recovery and hypertrophy under different dietary challenges.

Overall, our findings suggest that skeletal muscle retains a metabolic memory of endurance exercise, which enhances adaptations following RETRAINing. By comparing retrained mice to age-matched counterparts receiving the same exercise stimulus without previous exposure, we are able to control for exercise duration and volume, allowing us to assess the priming effect of past training. This memory of exercise appears to be encoded by mitochondrial pathways, facilitating oxidative metabolism and fatty acid utilization upon subsequent RETRAINing. Understanding how the mitochondria fuel long-term adaptations in skeletal muscle can provide a foundation for therapeutic interventions. Targeting this exercise memory to combat age-related declines in skeletal muscle mass and function could prove vital to an aging population. Given that mitochondrial dysfunction is a hallmark of sarcopenia and metabolic disease, identifying potential mechanisms to leverage the memory of exercise either in place of retraining or alongside it may offer potential approaches to maintaining muscle mass and health in aging populations.

Supplementary Material

Supplemental Figures S1–7, Table 1. https://doi.org/10.7910/DVN/OIOJ5U

DATA AVAILABILITY

Data will be made available upon reasonable request.